CFD-assisted heat-to-mass correction for open-vent boil-off prediction in an LN₂ benchmark tank

Converting total heat ingress directly into evaporation can overestimate open-vent boil-off because superheated vapor carries part of the incoming energy out of the tank. This study develops a high-fidelity computational fluid dynamics (CFD) framework and a reduced-order protocol for quasi-steady boil-off rate (BOR) prediction in a 1.5 m 3 Type-C liquid‑nitrogen tank. Numerical verification, conservation checks, model-sensitivity tests, and published BOR and temperature comparisons assess the multiphase conjugate model; BOR deviations remain below 0.55% at four filling ratios (FR) spanning 30–60%. Energy budgets over FR = 4–96% yield latent-conversion fractions of 0.6998–0.9561, correcting heat-based BOR overestimation of 42.9% and 4.6% at the respective endpoints. A filling-ratio correlation gives R 2 = 0.9992 and RMSE = 0.0022. The reduced-order protocol combines the correction with solid conduction, enthalpy-based vent-temperature estimation, and convective inner-wall boundaries. Its mean absolute relative BOR deviation is 0.32%, compared with 0.22% for CFD at the four matching conditions. It also retains the dominant insulation and saddle-guide heat-ingress contributions, supporting component-level thermal-load assessment. The reported online runtime decreases from 28.9 h to 0.1167 h on 96 CPU cores, approximately 248-fold, excluding construction of the conversion map. Intermediate cases excluded from fitting assess interpolation within the benchmark. A descriptor-based formulation provides a route to extension across cryogenic fluids, tank geometry and scale, and thermal environments; independent transferability and total predictive uncertainty remain to be established.